v2.3.0

NonlinearSolve v2.3.0

Diff since v2.2.1

Merged pull requests:

• WIP: Add DFSane method (#214) (@axla-io)
• docs: update advanced.md (#235) (@sathvikbhagavan)
• Impoving NLS Solvers (#236) (@avik-pal)
• Rename wrappers to be consistent with other SciML packages (#237) (@avik-pal)
• Add polyalgorithms and a default method (#238) (@ChrisRackauckas)
• Overhaul the documentation (#239) (@ChrisRackauckas)
• Change default in-place specialization (#240) (@ChrisRackauckas)
• update GPU tests (#242) (@ChrisRackauckas)
• Better Defaults: Auto AD Selection for Newton Methods (#244) (@avik-pal)
• Fix test to use TrustRegion (#246) (@ChrisRackauckas)
• Big docs update (#248) (@ChrisRackauckas)
• CompatHelper: add new compat entry for ModelingToolkit at version 8 for package docs, (keep existing compat) (#249) (@github-actions[bot])

Closed issues:

• Maxiters seems to be offset by 1 (#63)
• Error when using a neural network inside a system of nonlinear equations. (#114)
• Non-allocating stepping for TrustRegion (#122)
• Specialize Jacobians when doing Krylov methods to avoid building the matrices (#140)
• TrustRegion correctness (#142)
• Turn off trust region when the iteration is stuck in local minima (#146)
• More trust region schemes (#151)
• LinearAlgebra.SingularException(2) (#153)
• Solvers with KrylovJL fail (#158)
• Update README and docs tutorial examples to use new bracketing methods? (#209)
• Update docs for the halley method SimpleNonlinearSolve.jl (#175)
• Consider adding jacobian scaling (#180)
• Not finding a zero for a fairly simple problem (#187)
• :fixedpoint doesn't exist (#192)
• (SteadyState/Nonlinear)Problems from ODEProblems ignore kwargs (#204)
• Wrong solver in TrustRegion tests (#243)